Allelic polymorphism of GIGANTEA is responsible fornaturally occurring variation in circadian period inBrassica rapaQiguang Xiea,1, Ping Loua,1, Victor Hermanda, Rashid Amanb, Hee Jin Parkb, Dae-Jin Yunb, Woe Yeon Kimb,Matti Juhani Salmelac, Brent E. Ewersc, Cynthia Weinigc,d, Sarah L. Khana, D. Loring P. Schaiblea,and C. Robertson McClunga,2

aDepartment of Biological Sciences, Dartmouth College, Hanover, NH 03755-3563; bDivision of Applied Life Science (BK21 Program), Plant MolecularBiology and Biotechnology Research Center and Institute of Agriculture and Life Sciences, Gyeongsang National University, Jinju 660–701, South Korea;and Departments of cBotany and dMolecular Biology, University of Wyoming, Laramie, WY 82071

Edited by Susan S. Golden, University of California, San Diego, La Jolla, CA, and approved February 10, 2015 (received for review November 13, 2014)

GIGANTEA (GI) was originally identified by a late-flowering mu-tant in Arabidopsis, but subsequently has been shown to act incircadian period determination, light inhibition of hypocotyl elon-gation, and responses to multiple abiotic stresses, including toler-ance to high salt and cold (freezing) temperature. Genetic mappingand analysis of families of heterogeneous inbred lines showedthat natural variation in GI is responsible for a major quantitativetrait locus in circadian period in Brassica rapa. We confirmed thisconclusion by transgenic rescue of an Arabidopsis gi-201 loss offunction mutant. The two B. rapa GI alleles each fully rescued thedelayed flowering of Arabidopsis gi-201 but showed differentialrescue of perturbations in red light inhibition of hypocotyl elon-gation and altered cold and salt tolerance. The B. rapa R500 GIallele, which failed to rescue the hypocotyl and abiotic stress phe-notypes, disrupted circadian period determination in Arabidopsis.Analysis of chimeric B. rapa GI alleles identified the causal nucleo-tide polymorphism, which results in an amino acid substitution(S264A) between the two GI proteins. This polymorphism underliesvariation in circadian period, cold and salt tolerance, and red lightinhibition of hypocotyl elongation. Loss-of-function mutations ofB. rapa GI confer delayed flowering, perturbed circadian rhythmsin leaf movement, and increased freezing and increased salt toler-ance, consistent with effects of similar mutations in Arabidopsis.Collectively, these data suggest that allelic variation of GI—andpossibly of clock genes in general—offers an attractive target formolecular breeding for enhanced stress tolerance and potentiallyfor improved crop yield.

abiotic stress tolerance | circadian clock | hypocotyl elongation |flowering time | natural variation

The last half-century has seen dramatic increases in agricul-tural productivity. Despite the approximate doubling in

world population since 1964, the proportion with insufficientfood has dropped by ∼75%, although ∼1 billion remain un-derfed, and twice that many suffer from micronutrient defi-ciencies (1). Predicted growth in population and in per capitaconsumption will require an estimated doubling of crop pro-duction by 2050 (2). However, yield trends for maize, rice, wheat,and soybean—four major crops that currently produce nearlytwo-thirds of global agricultural calories—are insufficient toachieve this doubling (3). Therefore, there is a pressing need notsimply to sustain, but actually to accelerate yield improvement.One strategy to increase yield is to identify genetic variation in

plant regulatory networks that limit yield to define targets forprograms of marker-assisted (molecular) breeding. The circadianclock has been implicated as a target for increasing yield (4–6).Plant circadian clocks comprise multiple interlocked feedbackloops (7–9). There is natural variation in clock function in bothweedy and cultivated species (10–15), although few of the genesresponsible for these quantitative trait loci (QTL) have been

identified. We identified QTL for circadian period in a pop-ulation of Recombinant Inbred Lines (RIL) of Brassica rapa(14). Here we identify GIGANTEA (GI) as a major QTL re-sponsible for natural variation in circadian period and identifythe causal nucleotide polymorphism. We further show that thissame nucleotide polymorphism underlies variation in cold andsalt tolerance. We suggest that allelic variation of GI—andpossibly of clock genes in general—offers a tractable route formolecular breeding for enhanced stress tolerance and potentiallyfor improved crop yield.

ResultsB. rapa QTL for circadian period length were identified in a RILpopulation developed from a cross between the oilseed R500and the rapid cycling IMB211 (14). We exploited the referencegenomic sequence of B. rapa (16) to develop additional DNAmarkers to refine the map position of a period QTL, PERIODA9a(PERA9a), detected on chromosome A9 (Fig. 1 A and B) to aposition between two genes, Bra024534 and Bra024560 (Fig. 1C),and spanned by BAC B020D15 (17). Among the 27 genes in thatchromosomal region, GI (Bra024536) was a particularly strongcandidate to explain PERA9a (Fig. 1C) because GI affects circa-dian clock function in Arabidopsis (18–21).

Significance

The plant circadian clock affects many aspects of growth anddevelopment and influences both fitness in natural settings andperformance in cultivated conditions. We show that GIGANTEA(GI) underlies a major quantitative trait locus for circadian periodin Brassica rapa by fine-mapping, analysis of heterogeneous in-bred lines, and transgenic rescue of an Arabidopsis gi-201 loss-of-function mutant. Analysis of chimeric and mutated B. rapaGI alleles identified the causal nucleotide polymorphism re-sponsible for the allelic variation in circadian period, cold andsalt tolerance, and red light inhibition of hypocotyl elongation.Allelic variation of GI and of clock genes in general offers targetsfor marker-assisted (molecular) breeding for enhanced stresstolerance and potentially for improved crop yield.

Author contributions: Q.X., P.L., V.H., D.-J.Y., W.Y.K., B.E.E., C.W., and C.R.M. designedresearch; Q.X., P.L., V.H., R.A., H.J.P., M.J.S., S.L.K., and D.L.P.S. performed research; Q.X.,P.L., V.H., R.A., H.J.P., D.-J.Y., W.Y.K., M.J.S., B.E.E., C.W., and C.R.M. analyzed data; andQ.X., P.L., V.H., D.-J.Y., W.Y.K., B.E.E., C.W., and C.R.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Q.X. and P.L. contributed equally to this work.2To whom correspondence should be addressed. Email: c.robertson.mcclung@dartmouth.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1421803112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1421803112 PNAS | March 24, 2015 | vol. 112 | no. 12 | 3829–3834

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022

PERA9a was detected both at 12° and at 25 °C and at eachtemperature explained ∼10% of the variation in period length(14). To confirm the presence of PERA9a by genetic means, wetook advantage of residual heterozygosity in the GI region ofchromosome A9 in RIL256 (Fig. S1). We allowed RIL256 toself-fertilize and identified lines homozygous for IMB211(RIL256_IMB211) or for R500 (RIL256_R500) in the GI regionof A9 in an otherwise uniform genetic background. An addi-tional recombination during the development of RIL256_R500reduced the region of R500 DNA to ∼10 cM covering thePERA9a QTL (Fig. S1A). The period of RIL256_R500 wassignificantly shorter than that of RIL256_IMB211 (Fig. 1D andTable S1), which is consistent with the effects of the QTL (14).If GI were indeed the gene responsible for the PERA9a QTL,

it should be polymorphic in either sequence or expression levelbetween the RIL parents. We detected many single nucleotidepolymorphisms (SNPs) in the transcribed portions of the twoparental GI alleles (Fig. 2A). Most SNPs were predicted to befunctionally silent, either falling in introns or failing to changethe predicted amino acid sequence. However, three SNPs atnucleotide positions 1,274; 2,268; and 2,475 (numbered relativeto the A of the initiator AUG in a multiple sequence alignment)were predicted to result in the amino acid polymorphismsS264A, I541T, and D610V, respectively, where the amino acidfound in IMB211 is listed before and that found in R500 is listedafter the amino acid number. To test the functionality of the twoalleles, we performed transgenic rescue experiments in theArabidopsis gi-201 loss-of-function background, introducing fullgenomic copies of each B. rapa GI allele driven by their en-dogenous promoters. In our hands at 22 °C, gi-201 does not

affect circadian period (Fig. 2B, Fig. S2, and Table S1), con-sistent with earlier observations with gi-2, another strong loss-of-function allele (22). gi-201 mutants are late flowering (Fig.3A and Fig. S3A) and exhibit a long hypocotyl in red and bluelight (Fig. 3B and Fig. S3B). Both the IMB211 (BrGIIMB211) andR500 (BrGIR500) GI alleles fully rescued the late flowering de-fect of Arabidopsis gi-201 (Fig. 3A and Fig. S3A), indicating thatboth alleles are expressed and functional, at least for floweringtime. In contrast, BrGIIMB211, but not BrGIR500, fully rescued thelong hypocotyl in red light (Fig. 3B), indicating that BrGIR500 isdefective in this trait. Both BrGIIMB211 and BrGIR500 rescued thelong hypocotyl in blue light (Fig. S3B), although the rescueby BrGIR500 was only partial, suggesting that it is partially butnot fully functional for this trait. BrGIIMB211 had no effect on cir-cadian period of gi-201 but, strikingly, introduction of BrGIR500

resulted in a significant period shortening (Fig. 2B, Fig. S2A, andTable S1), consistent with the shorter period of R500 relative toIMB211. Both BrGI alleles show similar expression patterns inthe B. rapa RIL parents (Fig. S4A) and in Arabidopsis gi-201 (Fig.S4B), indicating that the differential effect on circadian perioddoes not result from differential expression of the two alleles.As noted above, three SNPs were predicted to change the

amino acid sequence of the GI protein. To determine whetherone or more of those changes was responsible for the functionaldifferences between the two alleles, we constructed chimericalleles, BrGIIMB211(R500) and BrGIR500(IMB211), in which a SacI–NdeI restriction fragment containing four SNPs at nucleo-tides 1,210; 1,213; 1,274; and 1,295 (Fig. 2A) was exchanged.

Fig. 1. Fine mapping and genetic definition of the PERA9a circadian periodQTL. (A) Circadian period QTL on chromosome 9 (PERA9a), redefined bycomposite interval mapping using the data of Lou et al. (14). (B) Fine map-ping of the PERA9a QTL using new molecular markers flanking the QTL.(C) Three overlapping BAC clones (17) spanning the PERA9a QTL with molec-ular markers used for fine mapping. (D) Period analysis of HIL RIL256_R500homozygous for R500 and RIL256_IMB211 homozygous for IMB211 in theregion of the PERA9a QTL. Box plots show median with the bottom and topof the box indicating 25% and 75%, respectively; whiskers indicate themaximum and minimum values.

Fig. 2. Transgenic complementation confirms GI as the gene underlying thePERA9a QTL. (A) Cartoon of the Arabidopsis and B. rapa GI genes with exonsrepresented by boxes (coding regions in black). Known gi mutations areindicated. The numbers 264, 541, and 610 indicate amino acid poly-morphisms detailed in Lower. Lower indicates the nucleotide and amino acidpolymorphisms between the R500, IMB211, and chimeric GI alleles. (B) Cir-cadian period of Arabidopsis Col-0 and gi-201 lines and of gi-201 lines car-rying the indicated B. rapa GI alleles. Box plots show median, with thebottom and top of the box indicating 25% and 75%, respectively; whiskersindicate maximum and minimum values. Different letters indicate valuesthat are statistically different as determined by ANOVA followed by Tukey’stest (Table S1).

3830 | www.pnas.org/cgi/doi/10.1073/pnas.1421803112 Xie et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022

When tested in transgenic plants, BrGIR500 shortened period,but BrGIR500(IMB211) did not (Fig. 2B, Fig. S2 A and B, and TableS1). Conversely, although BrGIIMB211 did not shorten period,BrGIIMB211(R500) shortened period to the same extent as BrGIR500

(Fig. 2B, Fig. S2 A and B, and Table S1). Thus, we conclude thatthe differential effects of these two alleles on period length areconferred by the polymorphisms in this SacI–NdeI restrictionfragment. Of the four SNPs in this region, only the one at nucleo-tide 1,274 was predicted to change the amino acid sequence.We therefore used site-directed mutagenesis to make two newGI alleles, BrGIIMB211(S264A) in which nucleotide 1,274 ofBrGIIMB211 was changed from T to G, changing amino acid 264from S to A, and BrGIR500(A264S) in which nucleotide 1,274 ofBrGIR500 was changed from G to T, changing amino acid 264from A to S. These alleles were introduced into Arabidopsisgi-201. Although BrGIR500 shortened period, BrGIR500(A264S) didnot (Fig. 2B, Fig. S2 A and C and Table S1). Conversely, althoughBrGIIMB211 did not shorten period, BrGIIMB211(S264A) shortenedperiod to the same extent as BrGIR500 (Fig. 2B, Fig. S2 A and C,and Table S1). All of the BrGI alleles were similarly expressed inArabidopsis gi-201 seedlings (Fig. S4 B and C), indicating that thedifferent effects on circadian period do not result from differ-ential expression of the transgenes. Thus, we conclude that thepolymorphism at nucleotide 1,274 (T1274G, encoding S264A,with the IMB211 allele listed first) is responsible for the differ-ential effects of these two alleles on period length and that it isthe presence of R500 information at nucleotide 1,274, encodingan A residue, that shortens period.We also tested the phenotypic consequences of introduction of

these chimeric and point mutant BrGI alleles on flowering timeand hypocotyl length. All six BrGI alleles rescued the late flow-ering time of Arabidopsis gi-201 equally (Fig. 3A and Fig. S3A).As indicated above, BrGIIMB211, but not BrGIR500, rescued thelong hypocotyl in red light (Fig. 3B). Both BrGIR500(IMB211) andBrGIR500(A264S), but neither BrGIIMB211(R500) nor BrGI IMB211(S26S4A),rescued the long hypocotyl in red light (Fig. 3B). In blue light, asimilar pattern was observed, although in this case BrGIR500,BrGIIMB211(R500), and BrGI IMB211(S26S4A) retained partial functionand partially but not fully rescued the long hypocotyl phenotypeof gi-201 (Fig. S3B). Thus, we conclude that the polymorphism atnucleotide 1,274 is responsible for the differential effects of thesealleles on hypocotyl length, and the presence of IMB211 infor-mation at nucleotide 1,274 is necessary to fully rescue the longhypocotyl in blue and red light phenotypes of gi-201.Our transgenic rescue experiments established the importance

of BrGI in circadian period and flowering time determination inArabidopsis. To test whether GI functions similarly in B. rapa, weidentified a set of TILLING mutations in B. rapa R-o-18, whichis closely related to R500 (23). Two putative null alleles, gi-1 andgi-3, were predicted to be strong loss-of-function alleles due tomissense mutations (gi-1, G1223A; W258Stop and gi-3, G1737A;W375Stop) and were analyzed after two back-crosses to the R-o-18parent. B. rapa lines homozygous for either gi-1 or -3 flowerlate (Fig. 4 A and B), consistent with loss of GI function. Cir-cadian period was unaffected at 18 °C (Fig. 4 C–E), but at 22 °C,most seedlings were arrhythmic, and those that showed circadianrhythms in leaf movement (31% of B. rapa gi-1 and 42% of gi-3mutant seedlings) showed long period and increased relativeamplitude error (RAE), a measure of the strength of a circadianrhythm (Fig. 4E and Table S1). Thus, mutational disruption ofGI weakens rhythms in B. rapa leaf movement at high temper-ature. This finding is consistent with Arabidopsis, where GI playsa critical role in maintaining rhythmicity in leaf movement athigher temperatures (24). However, the role of GI in rhythmicitymay not be limited to high temperature. In Arabidopsis, GI isimportant for maintenance of rhythmicity of CAB2:LUC ex-pression at both high and low temperatures (24).

Fig. 3. Transgenic complementation analysis of the role of GI in floweringtime, photomorphogenesis, cold tolerance, and salt tolerance. ArabidopsisCol-0 and gi-201 and gi-201 lines carrying the indicated B. rapa GI alleles weremeasured for flowering time, as measured by days to flowering (A); photo-morphogenesis, as indicated by hypocotyl length in continuous red light(cRL) (B); freezing tolerance, expressed as EL50, the temperature at which 50%of leaf electrolytes are lost (C); and salt tolerance, measured as fresh weightof total aerial tissues after growth on 1/2 strength MS medium containing15 mM NaCl relative to growth without NaCl (D). All data are presented asmean ± SEM. Different letters indicate values that are statistically differentas determined by ANOVA followed by Tukey’s test (Table S1).

Xie et al. PNAS | March 24, 2015 | vol. 112 | no. 12 | 3831

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022

Recently, GI has been implicated in cold (25), oxidative (26),drought (27), and salt (28) stress tolerance. Accordingly, wetested our panel of Arabidopsis gi-201 mutants carrying theB. rapa GI alleles for freezing tolerance, as measured by elec-trolyte leakage after exposure to freezing temperatures. gi-201showed significantly greater freezing tolerance than Col-0 (Fig.3C). Transformation of gi-201 with BrGIR500 or with either of thetwo BrGIIMB211 alleles carrying R500 information at nucleotide1,274 (BrGIIMB211(R500) or BrGI IMB211(S264A)) had no effect on

freezing tolerance (Fig. 3C). However, transformation of gi-201with BrGIIMB211 or with either of the two BrGIR500 alleles car-rying IMB211 information at nucleotide 1,274 (BrGIR500(IMB211)

or BrGIR500(A264S)) significantly reduced freezing tolerance, res-cuing the gi-201 phenotype of increased freezing tolerance (Fig.3C). Thus, we conclude that B. rapa GI plays a role in freezingtolerance and that the BrGIIMB211 allele is functional in this role,whereas the BrGIR500 allele has at least partially lost this func-tion, as has the Arabidopsis gi-201 allele. To confirm that GI alsoserves as a determinant of freezing tolerance in B. rapa, we testedB. rapa gi-1 and -3 mutants and found that loss of GI functionenhanced freezing tolerance, compared with the cognate wildtype, R-o-18 (Fig. 4F).Our data also suggest a role for GI in nitrogen accumulation.

gi-201 accumulates less N than does Col-0 (Fig. S3C). BrGIIMB211

and BrGIR500(IMB211) rescue the N content of gi-201, but BrGIR500

and BrGIIMB211(R500) do not. This finding suggests that IMB211information at nucleotide 1,274 is necessary for this rescue. How-ever, both point mutants (BrGI IMB211(S264A) and BrGIR500(A264S))rescue the N content of gi-201, which is inconsistent with the dif-ference between the two alleles being solely attributable to thatsingle position. Thus, regulation of leaf N content by allelic state atGI is not strongly supported by our data; further investigation willbe required to resolve the discrepancy among rescue lines.In contrast, we found no difference among our panel of Ara-

bidopsis gi-201 mutants carrying the B. rapa GI alleles for delta C13 (Fig. S3D) and rosette size (Fig. S3E). Thus, these alleles ofGI do not affect water use efficiency or plant size.In Arabidopsis, GI has been shown to function as a negative

regulator of resistance to salt stress, and gi loss-of-functionmutants show increased salt resistance (28). Accordingly, wetested our panel of Arabidopsis gi-201 mutants carrying theB. rapa GI alleles for salt tolerance as measured by fresh weightof aerial tissues following growth in the presence or absence ofNaCl. Growth of wild-type Col-0 seedlings or of glabrous-1 (gl-1)was reduced by ∼1/3 at 15 mM NaCl, but gi-201 was unaffected.Transformation of gi-201 with BrGIR500 or with BrGIIMB211(R500)

had no effect on the increased salt tolerance of gi-201 (Fig. 3Dand Fig. S5). However, transformation of gi-201 with BrGIIMB211

or with BrGIR500(IMB211) significantly reduced salt tolerance,rescuing the gi-201 phenotype of increased salt tolerance (Fig.3D and Fig. S5). Thus, we conclude that B. rapaGI plays a role insalt tolerance and that the BrGIIMB211 allele is fully functional inthis role, whereas the BrGIR500 allele is impaired in this function,as is the Arabidopsis gi-201 allele. To confirm that GI also servesas a determinant of resistance to salt stress in B. rapa, we testedB. rapa gi-1 and -3 mutants and found that loss of GI functionenhanced salt stress resistance compared with the cognate wildtype, R-o-18 (Fig. 4G).

DiscussionGI was first identified as a supervital mutant of Arabidopsis (29),with a prolonged phase of vegetative growth that increased re-productive capacity (30, 31). GI, in addition to this role in pho-toperiodic flowering, also is required for phytochrome signaling(20, 32, 33) and for proper clock function (18, 19), although theroles of GI in photoperiodic flowering, photomorphogenesis, andthe clock can be dissociated (20, 33, 34).Our results with the B. rapa GI alleles extend conclu-

sions drawn from Arabidopsis to Brassica. Both BrGIR500 andBrGIIMB211 fully rescue the photoperiodic flowering defect ofArabidopsis gi-201 mutants. gi-201 is a strong loss-of-functionallele, and full rescue of the gi-201 late-flowering phenotypedemonstrates that both Brassica alleles are equally functional interms of flowering time in Arabidopsis. However, the two B. rapaGI alleles are not equally functional in terms of hypocotylelongation, freezing tolerance, and salt tolerance. In each case,BrGIIMB211 fully rescued the gi-201 phenotype, whereas BrGIR500

Fig. 4. Characterization of B. rapa loss-of-function gi alleles. Putative loss-of-function gi alleles were identified by TILLING in B. rapa R-o-18 (23), whichis closely related to R500. (A) Images of gi-1 and -3 relative to wild-type R-o-18 taken as R-o-18 seeds are drying. (B) Quantification of flowering time ingi-1 and -3 relative to wild-type R-o-18 expressed as days to first flower.(C and D) Circadian pattern of cotyledon movement in B. rapa R-o-18, gi-1,and gi-3 at 22 °C (C) and 18 °C (D). (E) Quantification of circadian period ofR-o-18, gi-1, and gi-3 (mean ± SEM; n = 12–35) plotted vs. relative amplitudeerror (RAE), a measure of the strength of a circadian rhythm. An ideal cosinewave is defined as RAE = 0, and RAE = 1 defines the statistically detectablelimit of rhythmicity (53). (F) Effect of loss-of-function GI mutations onfreezing tolerance, expressed as EL50 (the temperature at which 50% of leafelectrolytes are lost) compared with isogenic wild-type B. rapa R-o-18.(G) Effect of loss-of-function GI mutations on salt tolerance compared withisogenic wild-type B. rapa R-o-18. Data are presented as the mean ± SEM ofshoot fresh weight of seedlings (n > 35) grown with 200 mM NaCl/averageshoot fresh weight of seedlings grown without NaCl. (B, F, and G) Differentletters indicate values that are statistically different as determined byANOVA followed by Tukey’s test (Table S1).

3832 | www.pnas.org/cgi/doi/10.1073/pnas.1421803112 Xie et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022

failed to rescue. Thus, the flowering timing function is distinctfrom the hypocotyl elongation, freezing tolerance, and salt tol-erance functions. Amino acid 264 (A in R500 vs. S in IMB211) istherefore important for these latter three functions. Amino acid264 is also important for circadian period determination, be-cause introduction of BrGIR500 (and BrGIIMB211 alleles carryingR500 information at amino acid 264), but not BrGIIMB211 (andBrGIR500 alleles carrying IMB211 information at amino acid264), shortens circadian period.GI functions typically involve protein–protein interactions. For

example, in flowering timing, GI interacts with the F-box proteinFLAVIN BINDING KELCH REPEAT F-BOX PROTEIN1(FKF1), and in the afternoon this GI–FKF1 complex degradesCYCLING DOF FACTORS (CDFs) bound at the CONSTANS(CO) promoter, relieving transcriptional repression and al-lowing the accumulation of CO mRNA in the light (35, 36). COprotein is stabilized in the light, accumulates, and induces ex-pression of FT, which induces floral identity genes. GI also bindsto the FT promoter and interacts directly with the FT repressors,SHORT VEGETATIVE PHASE (SVP), TEMPRANILLO 1(TEM1), and TEM2, to directly induce FT (37). Relevant to thenuclear roles of GI in the regulation of CO and FT expressionis a protein–protein interaction with ELF4 that confers sub-nuclear localization and restricts chromatin access of GI (38).Finally, GI protein stability is controlled by interaction withEARLY FLOWERING3 (ELF3) and the E3 ubiquitin ligaseCONSTITUTIVELY PHOTOMORPHOGENIC1 (COP1) (39).At least some of these functions are retained in B. rapa, becauseB. rapa gi loss-of-function mutants are late flowering. Both BrGIalleles fully rescue the flowering delay of Arabidopsis gi-201, sug-gesting that both BrGI proteins retain all these interactions.In its role in circadian period determination, in the light GI

binds to and stabilizes ZEITLUPE (ZTL), an F-box proteinclosely related to FKF1. The conformational shift in ZTL thatfollows the cessation of blue light signaling after dusk releasesGI, freeing ZTL to interact with and thereby target critical clocktranscriptional repressors, TOC1 and PRR5, for ubiquitylationand proteasomal degradation (40, 41). Because ZTL is a cyto-plasmic protein, the ZTL–GI interaction also retains GI in thecytosol, thereby limiting the nuclear accumulation of GI andantagonizing its roles in flowering timing and regulation of hy-pocotyl length (42).The effects on circadian period of BrGIR500 and BrGIIMB211

and their derivatives when transformed into Arabidopsis gi-201demonstrate an important role for amino acid 264 in circadianperiod definition. First, we note that BrGIR500 shortens circadianperiod length and acts as a gain-of-function mutant relative toBrGIIMB211. The amino acid change of S264 in BrGIIMB211 to Ain BrGIR500 is similar to that observed in the Arabidopsis gi-200short period allele (S932A) (20), although the two mutations liein different regions of the coding sequence. Moreover, theArabidopsis Col-0 amino acid at the position corresponding toB. rapa GI 264 is A, and this residue is conserved in four Ara-bidopsis accessions, barley and wheat (12). This finding suggeststhat the phenotypic consequences of differing information atamino acid 264 cannot be determined solely by the amino acid atthat residue but, rather, must be considered in the context of theentire protein sequence.One possible molecular explanation of the effect of the

BrGIR500 phenotype of short period could be an increased affinity ofBrGIR500 for ZTL, which would stabilize ZTL, increase ZTLaccumulation, and thereby shorten period (43). This explana-tion would be consistent with the observed effects on red lightinhibition of hypocotyl elongation, because the increased affinityof BrGIR500 for ZTL would limit nuclear accumulation andthereby antagonize the ability of BrGIR500 to rescue red lightinhibition of hypocotyl elongation. However, both B. rapa GIalleles fully rescue the flowering timing phenotype of gi-201.

Therefore, if differential affinities for ZTL explain the perioddifferences of the B. rapa GI alleles, these effects must not limitnuclear accumulation of GI to the point where flowering timingis compromised.The role of GI in red and blue light suppression of hypocotyl

elongation is incompletely defined. GI has been suggested toregulate CRY-mediated blue light signaling (44). The long hy-pocotyl in red light phenotype of gi-2 is suppressed in spindly-4(spy-4) gi-2 double mutants, and, consistent with this geneticallydefined interaction, SPY and GI proteins interact (22).More recently, GI has emerged as a key player in a number of

stress responses, notably to drought, cold, and salinity (25, 27,28). Loss of GI function in Arabidopsis results in increased tol-erance to both freezing and salt stress in Arabidopsis, and ouranalysis of B. rapa GI loss-of-function mutants shows a similarenhancement of freezing and salt resistance in B. rapa, suggest-ing that the roles of GI in resistance to these two stresses isconserved between species. Among these stresses, the role of GIin salt tolerance is best understood. In Arabidopsis, GI sequestersSALT OVERLY SENSITIVE2 (SOS2), a protein kinase thatserves as a positive regulator of salt tolerance (28). Release ofSOS2 from GI in response to elevated salt permits formation ofthe SOS2/SOS3 protein kinase complex that associates with andphosphorylates SOS1, activating its Na+/H+ antiport activity andenhancing salt tolerance (45). Differential rescue of the salttolerance phenotype of Arabidopsis gi-201 by BrGIIMB211 andBrGIR500allows the prediction of the Arabidopsis model is thatBrGIIMB211 has greater affinity for SOS2 than does BrGIR500.It is important to note that most of our experiments tested the

function of BrGI alleles in Arabidopsis. Extrapolating the com-plex network of GI functions elucidated in Arabidopsis intoB. rapa and determining the mechanistic basis of the functionaldifferences observed between the two BrGI alleles will likely becomplicated by the triplication of the B. rapa genome since itsseparation from Arabidopsis (46). In the B. rapa reference Chiifugenome, subsequent gene loss has eliminated two of the dupli-cated GI copies, leaving a single GI locus (47). Similarly, thereare single copies of SOS2 and of TEM2 (47). However, otherpotential GI interactors are present in two (CDF1, CDF2, CDF3,COP1, ELF3, FT, SPY, SVP, and TEM1) or three (ELF4) copies(47), and some or all of these copies have likely diverged fromtheir Arabidopsis homologs as well as from each other throughsubfunctionalization. Of interest in the context of circadian pe-riod determination, B. rapa has lost all copies of both ZTL andFKF1 (48). In B. rapa, the functions of ZTL and FKF1 are pre-sumably carried out by a third family member, LOV KELCHPROTEIN2 (LKP2), which exhibits partial functional redundancywith ZTL and FKF1 in Arabidopsis (49). In B. rapa, LKP2 ispresent in three linked copies resulting from a complex tandemgene triplication (48). It will be of interest to determine whethersubfunctionalization among these three copies of LKP2 has re-sulted in the specialization of one or more copies for floweringtiming or circadian functions.

Materials and MethodsPlant Materials. All constructs were transformed by floral dip into the Ara-bidopsis gi-201 mutant (20) carrying the proCCA1:LUC transgene (50). Seedswere sterilized in 20% (vol/vol) bleach and placed on half-strength Murashigeand Skoog (MS) medium with 0.8% agar and 2% (wt/vol) sucrose, thenstratified for 3 d at 4 °C in the dark.

Fine Mapping the GI QTL. To confirm the circadian period QTL on chromosome9, simple sequence repeat (SSR) markers were developed based on se-quenced BAC clones (17). In total, 13 SSR markers were used to fine-map theA9 QTL region (Table S2). Heterogeneous inbred lines (HILs) were generatedfrom an F4 RIL, RIL256, heterozygous in the A9 QTL region. We then geno-typed 144 plants of the F5 generation of RIL256 using our SSR markers toidentify HILs of RIL256 homozygous for either IMB211 (RIL256_IMB211) orR500 (RIL256_R500) in the QTL region.

Xie et al. PNAS | March 24, 2015 | vol. 112 | no. 12 | 3833

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022

Constructs. PCR products of full-length BrGIR500 and BrGIIMB211 genomic DNA,including promoters and 3′ UTR, were amplified from genomic DNA withprimer pairs Br_GI_locus_F1 and Br_GI_locus_R1 by using Phusion High-FidelityDNA Polymerase (New England Biolabs) and cloned into pENTR. SacI andNdeI were used to make chimeric genes, BrGIIMB211(R500) and BrGIR500(IMB211),in which the fragments that included the first different amino acid (S264A)were switched between BrGIIMB211 and BrGIR500, respectively. Three primers,I-S1F2, RI-S1R2, and R-S1F2, were used to make site-specific mutation con-structs, BrGIIMB211(S264A) and BrGIR500(A264S). All pENTR_GI constructs wererecombined into binary vector pH2GW7Δ (51).

Phenotypic Analysis. For circadian period determination, seedlings wereentrained in 12-h light/12-h dark LD cycles under white light (70 μmol·m2·s−1)at 22 °C before release into continuous light and temperature for LUC ac-

tivity measurement with a TopCount luminometer (Perkin-Elmer Life Sci-ences). Circadian period estimation in B. rapa lines was by cotyledonmovement as described (52). Data analysis was with BRASS (Version 2.1.4),which employs fast Fourier transform nonlinear least squares (53). Details ofother phenotypic analyses are provided in SI Materials and Methods. Sta-tistical significance for circadian period and other phenotypic analyses waswith ANOVA followed by Tukey’s test, which performs all pairwise com-parisons and corrects for multiple comparisons.

ACKNOWLEDGMENTS. We thank K. Greenham for helpful comments. Thiswork was supported by National Science Foundation Grants IOS-0923752and IOS-1025965, the National Research Foundation of Korea Grant2013R1A2A1A01005170, and Rural Development Administration, Republicof Korea Next-Generation BioGreen 21Programme, Systems & SyntheticAgrobiotech Center, Grants PJ01106901 and PJ01106904.

1. Godfray HCJ, et al. (2010) Food security: The challenge of feeding 9 billion people.Science 327(5967):812–818.

2. Tilman D, Balzer C, Hill J, Befort BL (2011) Global food demand and the sustainableintensification of agriculture. Proc Natl Acad Sci USA 108(50):20260–20264.

3. Ray DK, Mueller ND, West PC, Foley JA (2013) Yield trends are insufficient to doubleglobal crop production by 2050. PLoS ONE 8(6):e66428.

4. Dodd AN, et al. (2005) Plant circadian clocks increase photosynthesis, growth, survival,and competitive advantage. Science 309(5734):630–633.

5. Graf A, Schlereth A, Stitt M, Smith AM (2010) Circadian control of carbohydrateavailability for growth in Arabidopsis plants at night. Proc Natl Acad Sci USA 107(20):9458–9463.

6. Ng DW-K, et al. (2014) A role for CHH methylation in the parent-of-origin effect onaltered circadian rhythms and biomass heterosis in Arabidopsis intraspecific hybrids.Plant Cell 26(6):2430–2440.

7. McClung CR (2011) The genetics of plant clocks. Adv Genet 74:105–139.8. Hsu PY, Harmer SL (2014) Wheels within wheels: The plant circadian system. Trends

Plant Sci 19(4):240–249.9. McClung CR (2014) Wheels within wheels: New transcriptional feedback loops in the

Arabidopsis circadian clock. F1000Prime Rep 6:2.10. Michael TP, et al. (2003) Enhanced fitness conferred by naturally occurring variation

in the circadian clock. Science 302(5647):1049–1053.11. Edwards KD, et al. (2006) FLOWERING LOCUS C mediates natural variation in the

high-temperature response of the Arabidopsis circadian clock. Plant Cell 18(3):639–650.

12. Edwards KD, Lynn JR, Gyula P, Nagy F, Millar AJ (2005) Natural allelic variation in thetemperature-compensation mechanisms of the Arabidopsis thaliana circadian clock.Genetics 170(1):387–400.

13. Salathia N, Lynn JR, Millar AJ, King GJ (2007) Detection and resolution of genetic lociaffecting circadian period in Brassica oleracea. Theor Appl Genet 114(4):683–692.

14. Lou P, et al. (2011) Genetic architecture of the circadian clock and flowering time inBrassica rapa. Theor Appl Genet 123(3):397–409.

15. Anwer MU, et al. (May 27, 2014) Natural variation reveals that intracellular distri-bution of ELF3 protein is associated with function in the circadian clock. elife, 10.7554/eLife.02206#sthash.04mtZBPu.dpuf.

16. Wang X, et al.; Brassica rapa Genome Sequencing Project Consortium (2011) Thegenome of the mesopolyploid crop species Brassica rapa. Nat Genet 43(10):1035–1039.

17. Kim H, et al. (2009) Sequenced BAC anchored reference genetic map that reconcilesthe ten individual chromosomes of Brassica rapa. BMC Genomics 10:432.

18. Fowler S, et al. (1999) GIGANTEA: A circadian clock-controlled gene that regulatesphotoperiodic flowering in Arabidopsis and encodes a protein with several possiblemembrane-spanning domains. EMBO J 18(17):4679–4688.

19. Park DH, et al. (1999) Control of circadian rhythms and photoperiodic flowering bythe Arabidopsis GIGANTEA gene. Science 285(5433):1579–1582.

20. Martin-Tryon EL, Kreps JA, Harmer SL (2007) GIGANTEA acts in blue light signalingand has biochemically separable roles in circadian clock and flowering time regula-tion. Plant Physiol 143(1):473–486.

21. Dalchau N, et al. (2011) The circadian oscillator gene GIGANTEAmediates a long-termresponse of the Arabidopsis thaliana circadian clock to sucrose. Proc Natl Acad Sci USA108(12):5104–5109.

22. Tseng T-S, Salomé PA, McClung CR, Olszewski NE (2004) SPINDLY and GIGANTEAinteract and act in Arabidopsis thaliana pathways involved in light responses, flow-ering, and rhythms in cotyledon movements. Plant Cell 16(6):1550–1563.

23. Stephenson P, et al. (2010) A rich TILLING resource for studying gene function inBrassica rapa. BMC Plant Biol 10:62.

24. Gould PD, et al. (2006) The molecular basis of temperature compensation in theArabidopsis circadian clock. Plant Cell 18(5):1177–1187.

25. Cao S, Ye M, Jiang S (2005) Involvement of GIGANTEA gene in the regulation of thecold stress response in Arabidopsis. Plant Cell Rep 24(11):683–690.

26. Cao S, Jiang S, Zhang R (2006) The role of GIGANTEA gene in mediating the oxidativestress response and in Arabidopsis. Plant Growth Regul 48:261–270.

27. Riboni M, Galbiati M, Tonelli C, Conti L (2013) GIGANTEA enables drought escaperesponse via abscisic acid-dependent activation of the florigens and SUPPRESSOR OFOVEREXPRESSION OF CONSTANS. Plant Physiol 162(3):1706–1719.

28. Kim W-Y, et al. (2013) Release of SOS2 kinase from sequestration with GIGANTEAdetermines salt tolerance in Arabidopsis. Nat Commun 4:1352.

29. Rédei GP (1962) Supervital mutants of Arabidopsis. Genetics 47(4):443–460.30. Koornneef M, Hanhart CJ, van der Veen JH (1991) A genetic and physiological analysis

of late flowering mutants in Arabidopsis thaliana. Mol Gen Genet 229(1):57–66.31. Araki T, Komeda Y (1993) Analysis of the role of the late-flowering locus, GI, in the

flowering of Arabidopsis thaliana. Plant J 3:231–239.32. Huq E, Tepperman JM, Quail PH (2000) GIGANTEA is a nuclear protein involved in

phytochrome signaling in Arabidopsis. Proc Natl Acad Sci USA 97(17):9789–9794.33. Oliverio KA, et al. (2007) GIGANTEA regulates phytochrome A-mediated photomor-

phogenesis independently of its role in the circadian clock. Plant Physiol 144(1):495–502.

34. Mizoguchi T, et al. (2005) Distinct roles of GIGANTEA in promoting flowering andregulating circadian rhythms in Arabidopsis. Plant Cell 17(8):2255–2270.

35. Fornara F, et al. (2009) Arabidopsis DOF transcription factors act redundantly to re-duce CONSTANS expression and are essential for a photoperiodic flowering response.Dev Cell 17(1):75–86.

36. Sawa M, Nusinow DA, Kay SA, Imaizumi T (2007) FKF1 and GIGANTEA complex for-mation is required for day-length measurement in Arabidopsis. Science 318(5848):261–265.

37. Sawa M, Kay SA (2011) GIGANTEA directly activates Flowering Locus T in Arabidopsisthaliana. Proc Natl Acad Sci USA 108(28):11698–11703.

38. Kim Y, et al. (2013) ELF4 regulates GIGANTEA chromatin access through subnuclearsequestration. Cell Reports 3(3):671–677.

39. Yu J-W, et al. (2008) COP1 and ELF3 control circadian function and photoperiodicflowering by regulating GI stability. Mol Cell 32(5):617–630.

40. Más P, Kim W-Y, Somers DE, Kay SA (2003) Targeted degradation of TOC1 by ZTLmodulates circadian function in Arabidopsis thaliana. Nature 426(6966):567–570.

41. Fujiwara S, et al. (2008) Post-translational regulation of the Arabidopsis circadianclock through selective proteolysis and phosphorylation of pseudo-response regula-tor proteins. J Biol Chem 283(34):23073–23083.

42. Kim J, Geng R, Gallenstein RA, Somers DE (2013) The F-box protein ZEITLUPE controlsstability and nucleocytoplasmic partitioning of GIGANTEA. Development 140(19):4060–4069.

43. Somers DE, Kim WY, Geng R (2004) The F-box protein ZEITLUPE confers dosage-dependent control on the circadian clock, photomorphogenesis, and flowering time.Plant Cell 16(3):769–782.

44. Crepy M, Yanovsky MJ, Casal JJ (2007) Blue rhythms between GIGANTEA and phy-tochromes. Plant Signal Behav 2(6):530–532.

45. Quintero FJ, Ohta M, Shi H, Zhu JK, Pardo JM (2002) Reconstitution in yeast of theArabidopsis SOS signaling pathway for Na+ homeostasis. Proc Natl Acad Sci USA99(13):9061–9066.

46. Town CD, et al. (2006) Comparative genomics of Brassica oleracea and Arabidopsisthaliana reveal gene loss, fragmentation, and dispersal after polyploidy. Plant Cell18(6):1348–1359.

47. Cheng F, Wu J, Fang L, Wang X (2012) Syntenic gene analysis between Brassica rapaand other Brassicaceae species. Front Plant Sci 3:198.

48. Lou P, et al. (2012) Preferential retention of circadian clock genes during diploid-ization following whole genome triplication in Brassica rapa. Plant Cell 24(6):2415–2426.

49. Baudry A, et al. (2010) F-box proteins FKF1 and LKP2 act in concert with ZEITLUPE tocontrol Arabidopsis clock progression. Plant Cell 22(3):606–622.

50. Salomé PA, Xie Q, McClung CR (2008) Circadian timekeeping during early Arabidopsisdevelopment. Plant Physiol 147(3):1110–1125.

51. Karimi M, Inzé D, Depicker A (2002) GATEWAY vectors for Agrobacterium-mediatedplant transformation. Trends Plant Sci 7(5):193–195.

52. Xu X, Xie Q, McClung CR (2010) Robust circadian rhythms of gene expression inBrassica rapa tissue culture. Plant Physiol 153(2):841–850.

53. Plautz JD, et al. (1997) Quantitative analysis of Drosophila period gene transcriptionin living animals. J Biol Rhythms 12(3):204–217.

3834 | www.pnas.org/cgi/doi/10.1073/pnas.1421803112 Xie et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

15, 2

022